1
Plan for the IOCCG Atmospheric Correction Working Group
by Menghua Wang
Drafted on May 3, 2000
Revised on June 12, 2000
1. Goal of the Atmospheric Correction Working Group
The main objective of the atmospheric correction working group (WG) is to quantify the
performance of the various exiting atmospheric correction algorithms used in the various
ocean color satellite sensors. Therefore, the derived products from various ocean color
missions can be meaningfully compared and possibly merged. As the atmospheric correction
is a key procedure in the ocean color remote sensing, we want to answer question such as how
can derived ocean color products from one sensor be best compared with those from others.
We want to quantify the differences among the performance of the atmospheric correction
algorithms.
2. The Working Group Members
The core WG is composed as follows:
Antoine, Morel
MERIS
Deschamps
POLDER
Fukushima, Frouin OCTS/GLI
Gordon
MODIS
Wang
SeaWiFS
We can invite others to join us late on, e.g., after test data sets are generated and we start
to test our algorithms. A general requirement for people to join the WG is that they can
contribute a well documented atmospheric correction algorithm and participate at least some
of tests.
3. Atmospheric Correction Algorithms
The performance of the following atmospheric correction algorithms are required to be
tested and compared:
(1) SeaWiFS/MODIS algorithm (Gordon and Wang,1994)
(2) POLDER1/POLDER2 algorithm (Deschamps, POLDER L2 document, Feb. 1999)
(3) OCTS/GLI algorithm (Fukushima et al, 1998)
(4) MODIS Spectral Matching algorithm (Gordon et al, 1997; Chomko and
Gordon,1998)
(5) MERIS algorithm (Antoine and Morel, 1999)
After testing data are generated, we will invite others to test/compare their algorithms.
However, testing of the above 5 algorithms is the necessary requirement for the object of this
WG.
2
The derived parameters to be compared and tested are:
(a) normalized water-leaving reflectances;
(b) two-band ratio values of the derived normalized water-leaving reflectances;
(c) the atmospheric parameter--the derived aerosol optical thickness.
For comparison, the nominal center wavelengths for various ocean color sensors are
listed as following (selected bands which are close to SeaWiFS are listed):
SeaWiFS MODIS OCTS/GLI POLDER MERIS
-----------------------------------------------------------------412
412
412/412
---412
443
443
443/443
443
443
490
490
490/490
490
490
510
530
520/520
---510
555
550
565/565
565
560
670
670
670/680
670
665
765
750
765/749
765
708, 779
865
865
865/865
865
870
There are six bands in common (or close) to all sensors. The derived normalized waterleaving reflectances can therefore be compared for wavelengths at 443, 490, and 555/565 nm
as well as ratio values of 443/555 and 490/555 for all sensors. The aerosol optical thickness at
865 nm can also be tested.
4. Testing Data Sets
Test data sets are mainly from simulations with a RTE model for various cases. Some
data from SeaWiFS measurements are also considered as a complement to the simulated data
sets.
4.1 Simulated data sets
4.1.1 Open ocean cases
For the open ocean cases:
*
a polarized RTE code either with the successive-order-scattering (SOS) and/or MonteCarlo (MC) will be used;
*
TOA reflectances (Stokes I component) for spectral bands at nominal center wavelengths of 412, 443, 490, 555, 670, 708, 765, and 865 nm will be generated;
*
a two-layer plane-parallel atmospheric model, in which aerosols are located at the
bottom uniformly mixed with 22% of Rayleigh molecules, is proposed;
*
a Maritime aerosol model with RH=80% (M80) is used;
*
aerosol optical thicknesses at 865 nm as 0.05, 0.1, and 0.2 are used;
3
*
a Fresnel reflecting ocean surface with ocean pigment concentrations of 0.03, 0.1, 0.3,
and 1.0 (mg/m3) from model of Gordon et al. (1988) will be used;
*
TOA reflectances are generated with No gas absorptions (ozone, water-vapor, oxygen,
etc.) and No whitecap contributions;
*
solar and viewing geometry (in plane-parallel system):
solar angles of 0°, 45°, 60°, and 78°;
sensor viewing angles of 5°, 25°, and 45°;
relative azimuthal angle of 90°.
In addition, 1-2 cases with 12 solar-sensor viewing angle data set for the POLDER
tests will be generated.
Deschamps will provide me with specific POLDER solar-viewing gemoetry angles.
I would like to make a note that I don't expect we get into comparing various RTE
models/codes. Such comparisons will take tremendous efforts and a very long time.
4.1.2 Some cases for sensitivity studies
It is intended to have some sensitivity case studies including:
(a) sensor calibration errors--errors will be added into the TOA reflectances described in
section 4.1.1.
(b) absorbing aerosols--Urban aerosol model will be used for two type vertical distributions, i.e., two-layer and uniformly mixed one-layer, for the TOA data generations.
(c) case 2 water--although our algorithms mostly intended for case 1 water, we need a
quantitative estimation of atmospheric correction error over case 2 water.
Morel will provide 1-2 types of case 2 ocean contributions at sea surface for all 8
wavelengths.
4.1.3 Data from SeaWiFS measurements
As a complement to the simulated data sets, some data from SeaWiFS measurements representing various cases are proposed for the algorithm testing and comparison. For a useful
comparison, results from atmospheric corrections using two band pairs of both 670 & 865 and
765 & 865 will be tested.
(a) open ocean cases (with various locations and seasons);
(b) coastal region ocean waters;
(c) some trouble cases from SeaWiFS results, e.g., nLw<0, dust contamination, etc.
SIMBIOS project will help to collect and generate SeaWiFS data sets in which a colocated ocean in situ measurements are included. Therefore, we can have comparisons not
only among the atmospheric correction algorithms, but also with the in situ measurements.
4.2 Data format
4.2.1 Simulated data sets
4
Simulated data files (most likely in binary format) and a reading routine (Fortran) will be
distributed. The outputs from reading routine are: TOA reflectances at the 8 wavelengths
along with solar zenith angle, sensor viewing angle, and relative azimuthal angle. The relative
azimuthal angle is defined as that it is zero for the case of measurement of the direct sun glint
light from a flat ocean surface. These are inputs for the atmospheric correction algorithms.
4.2.2 Data from SeaWiFS measurements
We are all agreed that SeaDAS will be used to manipulate and process SeaWiFS data,
e.g., from SeaWiFS L1A to L1B.
5. Algorithm Testing/Comparison Procedure
(1) A small testing data file with a reading routine are planned to be first generated and
distributed within the WG. Therefore, individual member can start to work on their
algorithm, to make sure interface/coding/algorithm flow working properly.
(2) All simulated data sets are then distributed. Outputs from individual algorithm are
analyzed. Results are compared and shared within the WG. This is a very important stage.
Results from this give a basis for understanding the following testing and comparisons. We
therefore need spend some time to analyze/understand all results.
(3) Using results as a guideline from (2), data from SeaWiFS measurements can be
compared among the various atmospheric correction algorithms.
6. Time Schedule
(a) setting up working group: done.
(b) Draft a proposal for discussing in WG: May 16-18, 2000 (done).
(c) generate the test data sets: ~3-4 months.
(d) algorithm testing and results analyzing: 3-4 months.
(e) write up a IOCCG report: ~2-3 months.
(f) possible a journal paper?
References
Antoine, D., and A. Morel (1999), A multiple scattering algorithm for atmospheric correction
of remotely sensed ocean colour (MERIS instrument): principle and implementation for
atmospheres carrying various aerosols including absorbing ones, Int. J. Remote Sens., 20:
1875-1916.
Chomko, R., and H.R. Gordon (1998), Atmospheric correction of ocean color imagery: use of
the Junge power-law aerosol size distribution with variable refractive index to handle
aerosol absorption, Appl. Opt., 37: 5560-5572.
Deschamps et al.., (Feb. 1999), Descriptions of Level 2 ocean color products and algorithms,
POLDER document.
5
Fukushima, H., A. Higurashi, Y. Mitomi, T. Nakajima, T. Noguchi, T. Tanaka, and M.
Toratani (1998), Correction of atmospheric effects on ADEOS/OCTS ocean color data:
Algorithm description and evaluation of its performance, J. Oceanogr., 54: 417-430.
Gordon, H.R., O.B. Brown, R.H. Evans, J.W. Brown, R.C. Smith, K.S. Baker, and D.K. Clark
(1988), A semianalytic radiance model of ocean color, J. Geophys. Res., 93: 10909-10924.
Gordon, H.R., and M. Wang (1994), Retrieval of water-leaving radiance and aerosol optical
thickness over the oceans with SeaWiFS: A preliminary algorithm, Appl. Opt., 33: 443452.
Gordon, H.R., T. Du, and T. Zhang (1997), Remote sensing of ocean color and aerosol
properties: resolving the issue of aerosol absorption, Appl. Opt., 36: 8670-8684.